AI模型精选76°

Nvidia 即将发布 Nemotron 3 Ultra,混合 SSM+MoE 架构

Nemotron 3 Ultra will be available from Nvidia in …

精选理由

Nemotron 3 Ultra 的混合 SSM+MoE 架构解决了长序列推理的高成本痛点,做长上下文应用或工具调用的开发者值得关注,可以直接期待其发布。

AI 摘要

Nvidia 将在几天内发布 Nemotron 3 Ultra 模型。该模型采用混合状态空间模型(SSM)与混合专家(MoE)架构,SSM 部分专为长序列设计,使模型能够更长时间地保持推理或使用工具,而不会被常规注意力机制的计算成本压垮。黄仁勋在 NVIDIA GTC Taipei 2026 上宣布了此消息。

原文 · rohanpaul_ai

Nemotron 3 Ultra will be available from Nvidia in …

Nemotron 3 Ultra will be available from Nvidia in few days.

Hybrid SSM (state-space models) + mixture-of-experts architecture.

The SSM part is built for long sequences, so the model can keep reasoning or using tools for longer without getting crushed by the usual attention cost.

Jensen Huang at NVIDIA GTC Taipei 2026

----

From 'NVIDIA' YT channel (link in comment)